Leader of the group Interactive Information Access  (INS2): Arjen de Vries

The Interactive Information Access (INS2) research group investigates how access to information can be improved with humans ‘in the loop’, in the context of the rich heterogeneous information collections that arise as dataspaces; a relatively new concept introduced in the database field that simply refers to all data in scope of an entity (an organisation, a person, but also a smart home).

From a bottom-up, data-driven perspective, we aim to understand how an information system’s internal representations should evolve based on observed interactions: a mix of implicit user annotations (e.g., query-click pairs) and explicit ones, both automatically derived descriptions using computer vision and natural language processing as well as manually provided annotations, including traditional metadata as well as tags and mentions in social media.

From a top-down perspective, the question is how to design key abstractions in the user interface that maximize the support of the end user's task while minimizing the assumptions about the (structure of) the heterogeneous data.

In both perspectives, entities play a central role; they provide relatively unambiguous pieces of information that may play a role as anchors between the user’s information need and the representations available to the system.

The group is part of the cluster Information Systems (INS).